Intended for healthcare professionals

Editorials

Third generation oral contraceptives

BMJ 2000; 321 doi: https://doi.org/10.1136/bmj.321.7255.190 (Published 22 July 2000) Cite this as: BMJ 2000;321:190

Caution is still justified

  1. David C G Skegg, professor
  1. Department of Preventive and Social Medicine, University of Otago, PO Box 913, Dunedin, New Zealand

    Oral contraception is effective, convenient, and reversible. For most women it is also remarkably safe. Some people involved in family planning go further and imply that the oral contraceptive pill is almost free of risk. This creates an illusion that is shattered whenever adverse effects—however rare—are brought to light. In October 1995 the Committee on Safety of Medicines in the United Kingdom warned that oral contraceptives containing desogestrel or gestodene carried a small increase in risk of venous thromboembolism compared with older preparations. The chaos that followed in Britain (and a few other countries) may have stemmed partly from an illusion of absolute safety, as well as from the publicity triggered by the warning.

    The committee's announcement was followed by the publication of four well designed studies that gave a consistent picture: women using third generation oral contraceptives containing desogestrel or gestodene had about twice the risk of venous thromboembolism of women using preparations containing levonorgestrel.1 In the four years that have followed, a plethora of articles, reviews, and symposiums have implied that these reports were flawed and that reanalyses or new studies show no differences in the risk of venous thromboembolism. The debate has been not only acrimonious but also confusing to many doctors unfamiliar with the arcane techniques of pharmacoepidemiology.

    In fact the authors of three of the four studies stand by their original conclusions. The exception is the case-control study known as the transnational study, which has been subjected to a dizzying series of reanalyses.2 In 1996 the authors reported that the risk of venous thromboembolism was higher for oral contraceptives marketed more recently; this was taken as evidence of bias due to “attrition of susceptibles” (the supposition being that women who might have developed this adverse event with older contraceptives would already have done so).3 The new analysis focused on one age group, and the pattern was not seen overall. 1 4 A second reanalysis used quadratic spline modelling, a statistical technique for smoothing risk estimates over time, and found that first time users of second and third generation oral contraceptives had the same risk.5 Later evidence showed that this interpretation was probably an artefact due to constraints placed on the spline model.4

    Should interruption periods be adjusted for in the analysis?

    A third attempt used information about past use of oral contraceptives but then analysed the data as if they had arisen from a cohort study.6 The validity of the analysis and conclusions has been challenged.4 A fourth revision is reported to show that the observed difference in the risk of venous thromboembolism with second and third generation contraceptives might be due largely to variations in the duration and patterns of use.7 The analysis is restricted to a minority of women who had interrupted oral contraception in the past, and it is unclear that adjusting for the duration of interruption periods is appropriate.

    In one of the original four studies, Jick et al used the British general practice research database.8 This source of data has also been analysed more than once by Farmer and colleagues.9 Their adjusted analyses show no material differences in risk between oral contraceptive formulations, but the reasons for the discrepancy from the original study have not been adequately explained. Possible factors may be that Jick's group used stricter criteria for inclusion of practices and for identification of women with venous thromboembolism, which is notoriously difficult to diagnose. Meanwhile the original group has published two further studies from the same database that support the difference between second and third generation contraceptives. 10 11

    That two groups using the same computerised database (though with overlapping rather than identical data) report opposite conclusions must be of concern to all who rely on the results of pharmacoepidemiology. Equally perplexing is the observation that of nine reported studies without pharmaceutical industry funding, one found no difference and the other eight found relative risks from 1.5 to 4.0 (summary relative risk for all nine studies 2.4), while four sponsored studies found relative risks between 0.8 and 1.5 (summary relative risk 1.1).12

    The rate of activated protein C

    A randomised crossover trial has confirmed that acquired resistance to activated protein C is more pronounced during use of an oral contraceptive containing desogestrel than one containing levonorgestrel.13 Whether this completely explains the higher risk of venous thromboembolism with third generation oral contraceptives is uncertain, but we can no longer assert that there is no plausible biological mechanism for their increased thrombogenicity. Although randomised trials of the incidence of venous thromboembolism will never be feasible, the evidence from most observational studies suggests that there is a genuine difference in risk between third generation oral contraceptives and older preparations. The alternative explanations invoking bias or confounding do not stand up to quantitative scrutiny. 1 4

    The absolute risks are small, but are they of no consequence? Studies in Britain and New Zealand, where third generation pills have been commonly used, estimate that the annual death rate from idiopathic venous thromboembolism in users of oral contraceptives is about 1 in 100 000. 9 14 Thus the risk of dying of a woman using the pill for two or three years is of the same order of magnitude as the risk of fatal aplastic anaemia in a patient treated with chloramphenicol.15 Non-fatal events, which can still have serious consequences, may be at least 30 times more common.9

    Many women will be interested in the possibility of halving such risks by choosing one of the pills that have been standard in most countries (including the United States and Australia). If they encounter side effects that might be diminished or avoided by switching to a third generation preparation, they may well feel that the small extra risk is worth taking.

    References

    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    8. 8.
    9. 9.
    10. 10.
    11. 11.
    12. 12.
    13. 13.
    14. 14.
    15. 15.